Dielectric Modification of 5V‐Class Cathodes for High‐Voltage All‐Solid‐State Lithium Batteries

A “5V-class” all-solid-state lithium battery (Li/Li3.2PO3.8N0.2/LiCr0.05Ni0.45Mn1.5O4-δ) demonstrates an improved rate capability when its electrolyte/cathode interface is modified by dielectric BaTiO3 nanoparticles. Such “dielectric modification” is supposed to be able to resolve a Li+-deficient layer at the interface, which has limited the charge-transfer reactions rates. There is an urgent demand to popularize high-energy and high-power rechargeable batteries with no safety concerns for automobile applications; “5V-class” all-solid-state lithium batteries are one of the solutions. Here we demonstrate that the charge-transfer reaction rate can be enhanced by four orders of magnitude at solid electrolyte (LiPON)/5V-class cathode (LiCr0.05Ni0.45Mn1.5O4-δ) interface when the electric potential distribution at the interface is designed by dielectric BaTiO3 nanoparticles (BTNs). The resultant BTN-modified Li/LiPON/LiCr0.05Ni0.45Mn1.5O4-δ battery demonstrated an improved rate capability with discharge capacity of 100 mA h g−1 at 8C rate. The ever-increasing demand for electric vehicles (EVs) is aspiring for next-generation rechargeable batteries with higher energy densities. Because the energy density of a battery (W h L−1) is defined as the product of its capacity (A h L−1) and voltage (V), the increase in its voltage is a conclusive approach to improve its energy density. The operating voltage of a battery is determined by the difference in electrode potentials between cathodes and anodes, both of which principally need to be within the potential stability windows of electrolytes in order to avoid their electrochemical decomposition over prolonged charge/discharge cycles. Therefore, high-voltage batteries cannot be realized without electrolytes with wide potential stability windows. In fact, looking back on the history of rechargeable batteries from the standpoint of their operating voltage, Ni-Cd and Ni-MH batteries1 are classified as the “first-generation” because their voltages are limited as small as 1.2 V, mainly due to the narrow potential stability windows of aqueous-based electrolytes. Indeed Ni-MH batteries have yielded great commercial success as power sources for hybrid vehicles (HVs) since the release of the Prius in 1997, but their limited energy density has restricted their application for pure EVs that require more energy than HVs. The low energy densities of Ni-MH batteries was overcome by the “second-generation” lithium-ion batteries operating at ca. 4 V range using 4V-class cathodes such as LiCoO2,2, 3 LiMn2O4,4 Li(Ni, Co, Al)O2,5 or Li(Ni, Co, Mn)O2.6 Their high-voltage is owed mainly to the wide potential stability windows of non-aqueous solvents such as propylene carbonate (PC), ethylene carbonate (EC), diethyl carbonate (DEC), and dimethyl carbonate (DMC), etc.,7 in liquid electrolytes.8 Although such 4V-class lithium-ion batteries are now going to be dominantly used as power sources for EVs, a strong demand to improve their driving range is awaiting the realization of the “third-generation” batteries operating at ca. 5 V range. To date, several kinds of 5V-class cathodes,9 such as LiNi0.5Mn1.5­O4-δ,10 Cr-doped LiNi0.5Mn1.5O4,11 LiCoPO4,12 Li2CoPO4F,13 LiNiPO4,14, 15 Li2NiPO4F,16, 17 LiNiVO4,18, 19 etc. have been reported. However, when they are used with conventional liquid-based electrolytes, they face problems in terms of durability and cycle stability because they give rise to strong oxidation atmospheres against the electrolytes and decompose them.20 Although the kinetics of such side-reactions can be slowed down by techniques like surface coating on cathode materials,21 concentration-gradient cathodes,22 and additives into the electrolyte,23, 24 etc., it should not be possible to suppress the side-reactions completely as long as liquid-based electrolytes are used. Such an underlying problem in the 5V-class batteries led us to utilize inorganic solid electrolytes8 that generally possess wider electrochemical potential windows compared to conventional liquid-based electrolytes. The resultant 5V-class all-solid-state lithium batteries are expected to ensure not only better durability, but also improved safety because of the non-flammable properties of the solid electrolytes.25, 26 As an example of an inorganic solid electrolyte, in this work we used a lithium-oxynitride phosphate glass, Li3.2PO3.8N0.2 (LiPON), that is reported to be stable up to 5.5 V (vs. Li/Li+).27 The LiPON was combined with 5V-class LiCr0.05Ni0.45Mn1.5O4-δ (LNM) cathodes11 with spinel structure whose charge/discharge capacities are predominantly given at around 4.8 V (vs. Li/Li+) using the Ni2+ ⇌ Ni4+ redox couple as well as small capacities at around 4.0 V (vs. Li/Li+) using the Mn3+ ⇌ Mn4+ redox couple. The as-prepared 5V-class Li/LiPON/LNM battery, however, exhibited no charge/discharge reactions at all at potentials between 3.0 and 5.3 V in potential sweep measurements carried out at 0.1 mV s−1. A.c. impedance analysis revealed that its poor electrochemical performance was attributed to the huge charge transfer resistance at the LiPON/LNM interface (RLiPON/LNM > 107 Ω cm2) measured at 4.7 V where reversible charge/discharge reactions should have occurred. Such a poor performance in the 5V-class battery was unexpected to the authors because we have previously observed reasonable charge/discharge reactions in 4V-class all-solid-state batteries (Li/LiPON/LiCoO228 and Li/LiPON/LiMn2O429). One of the possible causes to yield the huge RLiPON/LNM in the 5V-class Li/LiPON/LNM battery is the large electric field at the interface (ELiPON/LNM) generated by the large electric potential difference between the two materials (φLNM – φLiPON). Therefore, with a view to reduce the large ELiPON/LNM, we attempted to modify the LiPON/LNM interface by dielectric materials of BaTiO3, and studied how the RLiPON/LNM was influenced by the modification. BaTiO3 nanoparticles (BTNs) with 10 or 100 nm in diameters were modified at the LiPON/LNM interfaces. It is known that the dielectric constants of BTNs strongly depend on their particle sizes: their dielectric constants are reported to be ca. 2 × 102 and ca. 3 × 103, respectively.30 Figure 1a illustrates a thin-film battery studied in this work. BTNs (d = 10 nm) were deposited on LNM cathodes by spin-coating of precursor liquids containing BTNs in 2-methoxy ethanol solvent (JGC C the coverage ratio of the BTNs was estimated as 24% by comparing the intensities of Ba and Ni in X-ray photoelectron spectroscopy (XPS). It is noteworthy that no capacity degradation was observed in the optimal “0.008 wt% battery” after 100 charge/discharge cycles at 2C rate between 3.5–4.9 V, see Figure 2. The remarkable cycle stability is owed not only to the electrochemical stability of the LiPON at the high voltage, but also its elastic property that can adapt the volume change in the LNM cathode after continuous charge/discharge cycles. We also examined the size effect of the BTNs on the battery performances. Figure 3a represents a series of Nyquist plots of BTN (d = 100 nm)-modified Li/LiPON/LNM batteries. The semicircles in Figure 3a correspond to charge-transfer resistances at the LiPON/LNM interfaces (RLiPON/LNM) measured at 4.7 V. The RLiPON/LNM decreased as the amount of BTNs (d = 100 nm) increased. The optimal RLiPON/LNM of 2 × 103 Ω cm2 was recorded in a battery where BTNs (d = 100 nm) were deposited on the LNM cathode by the electrospray deposition for 120 min. The optimal RLiPON/LNM was about 4 orders of magnitude smaller than that in the unmodified battery (RLiPON/LNM > 1 × 107 Ω cm2). As seen in the FE-SEM images in Figure 3a, the optimal battery possesses BTNs (d = 100 nm) at intervals of every 1–2 μm, which is greater than 30-50 nm in the case of BTNs (d = 10 nm) as has been proved in Figure 1. This result suggests that the optimal interval of BTNs should depend on their dielectric constants; that is, BTNs (d = 100 nm) are supposed to have a more significant effect in reducing RLiPON/LNM because they possess ca. 15-times larger dielectric constant than BTNs (d = 10 nm). Figure 3b represents a rate capability of the BTN (d = 100 nm)-modified Li/LiPON/LNM battery in comparison with that of an unmodified battery. When the BTN (d = 100 nm)-modified battery was discharged at 8C rate, it retained 84% of its discharge capacity of 120 mA h g−1 recorded at 0.25C rate. It should be also noted that no capacity degradation was observed over 20 charge/discharge cycles in the BTN-modified battery. In contrast, the unmodified battery showed only 45 mA h g−1 at 0.25C rate. Hereafter we discuss the effect of BTNs on improving the charge transfer rates at the LiPON/LNM interface. In general, 5V-class batteries possess larger electric field at their electrolyte/cathode interfaces than 4V-class batteries due to the large potential difference between the two materials. Such a large interfacial electric field in the 5V-class battery (ELiPON/LNM) becomes a driving force to extract a lot of lithium ions from LiPON to LNM at open-circuit conditions, as illustrated in Figure 4a. As a result, Li+-deficient layers should be developed at the LiPON-side of the interface, which in turn behave as a resistive layer for charge-transfer reactions. Such a phenomenon was previously discussed by Takada and co-workers31-35 based on the space-charge theory.36 In contrast, when dielectric BTNs are modified at the LiPON/LNM interfaces, as illustrated in Figure 4b, electric dipoles in the BTNs should be arranged in such a way to reduce the large ELiPON/LNM: their negative charges will face the positive charges in the LNM cathode, whereas their positive charges will face the other side. As a result, as indicated by the dashed gray arrows in Figure 4b, lithium ions originally located behind the BTNs are expected to migrate toward the vicinities of the LNM/LiPON/BTN triple-phase-boundaries so as to maintain local charge neutralities. Such a rearrangement of the Li+ distribution will yield “Li+ pathways” for charge transfer reactions as indicated by double-headed arrows in Figure 4b, where the Li+ concentration in LiPON is approximately equal to that in the stoichiometric Li3.2PO3.8N0.2. Such “Li+ pathways” are supposed to spread concentrically away from the centers of BTNs, the radii of which should depend on their sizes (i.e., dielectric constants): it would appear that the BTNs (d = 10 nm) and BTNs (d = 100 nm) yield “Li+ pathways” concentrically with radii of ca. several tens of nanometers and ca. 1–2 μm, respectively, according to the experimental results in Figure 1 and 3. In contrast, when TiO2 (d = 100 nm) nanoparticles were modified at LiPON/LNM interfaces as a comparison experiment, the battery showed no charge/discharge reaction at all (Figure S1 in the Supporting information), because TiO2 (d = 100 nm) possess only ca. 1/30 smaller dielectric constants than that of the BTNs (d = 100 nm). It is expected that the role of dielectric properties in reducing RLiPON/LNM will be clarified in more detail using advanced analytical techniques like electron holography (EH)-equipped transmission electron microscopy (TEM)37 that can analyze electric potential distributions at around electrode/electrolyte interfaces, together with the aid of modelling studies on electric potential distribution and the Li+ concentration profile around the interface. In summary, we have studied the impact of electric potential distributions at electrolyte/cathode interfaces on charge-transfer reactions in 5V-class all-solid-state batteries. It was revealed that an as-prepared 5V-class Li/LiPON/LNM battery had huge charge-transfer resistance (RLiPON/LNM > 1 × 107 Ω cm2), which would be due to the greatly developed Li+-deficient layers at the LiPON-side of the interface that were caused by the large electric potential difference (φLNM – φLiPON). The large RLiPON/LNM could be reduced down to 2 × 103 Ω cm2 by designing the electric potential distribution around the interface using dielectric materials of BTNs. Such “dielectric modification” induces rearrangement of the Li+ distribution at the interface, and eventually yields “Li+ pathways” for charge-transfer reactions. It should be noted that the volume fraction of the modified BTNs is only ca. 0.04% compared to the total volume of the battery; therefore the “dielectric modification” addressed in this work does not lose energy density of batteries. Nevertheless, the charge-transfer reaction rate could be enhanced by 4 orders of magnitude. It is concluded that dielectric materials (e.g., BaTiO3) can be used as modifying agents at electrolyte/cathode interfaces in 5V-class all-solid-state lithium batteries with a view to increasing their power densities without losing their energy densities. Preparation of a BTN-modified Li/LiPON/LNM Battery: A crystalline LNM thin film (60 nm in thickness) was deposited on a Pt-coated glassy carbon substrate (973 K) by pulsed laser deposition (PLD) with light source of a fourth-harmonic neodymium-doped yttrium aluminum garnet (Nd:YAG) laser (266 nm; LOTIS TII LS-2137/4-BBO) with fixed energy fluence (1.4 J cm−2) and fixed repetition frequency (10 Hz). The oxygen partial pressure inside the chamber was 27 Pa. BTNs (d = 100 nm) were dispersed on the LNM film by electrospray deposition (PDS-D, HAMAMATSU Nano Technology Inc.), where precursor liquids contained BTNs (d = 100 nm) dispersed in 2-methoxy ethanol solvent with solid content concentration of 0.08 wt%. The diameter of the nozzle was 24 μm and the distance between the nozzle and substrate was 20 mm. The applied voltage between the nozzle and the substrate was 1450 V. Meanwhile, BTNs (d = 10 nm) were dispersed on LNM films by spin-coating, as described in the main text. Then LiPON thin films (2.5 μm in thickness) were deposited on the BTN-dispersed LNM thin films by RF magnetron sputtering as reported elsewhere.27 The resultant LiPON/LNM laminate was annealed at 498 K in air. Finally, a lithium anode (1.0 μm in thickness) was deposited on the LiPON film by vacuum evaporation. The geometric electrode area of the battery was 0.20 cm2. Electrochemical Characterization of the Batteries: The batteries were characterized by the potential sweep method (0.1 mV s−1) in the range of 3.0–5.3 V, galvanostatic charge/discharge tests between 3.5–4.9 V at 0.25, 1, 2, 4, 8 C rate, as well as a.c. impedance spectroscopy at 4.7 V in the frequency of 500 kHz to 0.1 Hz with amplitude of 10 mV. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.